Parent Directory
| 
Revision Log
| Links to HEAD: | (view) (annotate) |
| Sticky Revision: |
- res.c:query_name() remove memset that's not needed
- res.c: rename reslist::namelength to reslist::name_len
- res.c: remove AR_TTL definition we don't need
- Correctly prefix all functions found in reslib.c with reslib_ instead of irc_
- Bump copyright years
- res.c:res_readreply(): remove pointless 'continue'
- res.c: remove timeout_resolver() wrapper; remove unused 'next_time' calculation in timeout_query_list() as this function is run each second anyway
- Bump copyright years
- Cleanup ip address matching routine in various places
- Stylistic changes
- Bump copyright years everywhere
- res.c:do_query_number(): kill another memcpy()
- Add support for monotonic clocks and have some timers make use of it; remove event_set_back_events()
- Add address_compare() and make use of it in some places
- Update copyright years
- Make use of the bool data type in some more places
- Style corrections
- Stylistic changes
- Rip out mempool
- res.c:restart_resolver(): Added sanity test. ResolverFileDescriptor might be NULL at this point
- Restore fd_table. No longer allocate fde_t items from within any other structures like the AuthRequest, or Connection structure - struct AuthRequest once again is no longer allocated from within the Connection structure
- Update copyright years
- Update copyright years
- Style corrections
- Style corrections in several places
- Fixed svn properties
- res.c: minor stylistic changes
- res.c: reslist::id is now of unsigned type; remove reslist::sent which we currently don't need
- res.c:send_res_msg(): remove extraneous parentheses
- Rename init_resolver() to resolver_init()
- Now that we got time_t to work nicely on openbsd with snprintf's conversion specifiers, we ran into a similiar issue on Raspbian/ARMv7's time_t which is of signed 32 bit and doesn't cope at all with %j. Instead of doing tricks, get rid of time_t everywhere and forever and use uintmax_t instead which has at least a 'standardized' conversion specifier associated with it.
- Update copyright years
- res.c: remove unused header includes
- Move report_dns_servers() from res.c to m_stats.c
- Update copyright years
- Fixed compile warnings
- Renamed variables
- Update GPL 2 license headers
- res.c: style corrections
- res.c:query_name(): get rid of useless typecasting
- res.c:proc_answer(): error out and leave the client unresolved if we get an unexpected record type
- Removed stupid PF typedef in fdlist.h which prevented both gcc and clang from spitting out a warning about the first argument of s_bsd:ssl_handshake() being an 'int' when it really has to be a 'fde_t' pointer. - Fixed first argument of s_bsd:ssl_handshake() which should be a 'fde_t' pointer instead of an 'int'.
- conf.c:conf_dns_lookup(): use gethost_byname_type() - res.h, res.c: removed gethost_byname()
- res.c:proc_answer(): delete useless castings
- res.c:proc_answer(): use 'unsigned'
- res.c: removed "request_state" enum
- res.c:proc_answer(): applied "Ignore CNAME records in dns queries" fix from p4.
** Ignore CNAME records in dns queries. Allows PTRS to point to CNAMEs
which can point to an A/AAAA which resolves to the correct IP.
- res.c: removed "try A if AAAA failed" logic
- res.c:start_resolver(): made /stats F more descriptive about the resolver socket
- Renamed various variables
- res.c: was missing +1 in reslist::name
- res.c:init_resolver(): removed useless memset()
- Removed ipv6 detection. We now assume all systems that run hybrid have ipv6 availability and sockaddr_storage.
- Fixed resolver/auth issues with too long hostnames
- res.c:res_readreply(): read as many packets as possible from kernel buffer in a single run. This might increase performance for heavy loaded servers doing lots of dns requests. As side-effect it fixes issues with edge-triggered epoll (we're only doing level-triggered right now)
- res.c:res_readreply(): don't grant further retries if the packet couldn't be decoded. We only give it one shot.
- res.c: reformatting; style corrections; updated comments
- res.c:res_readreply(): use ssize_t for 'rc'
- Fixed typos all over the place
- res.c:res_readreply(): removed sparc/alpha hack
- res.c:res_readreply(): don't do anything until we checked for possible fake replies
- res.c:res_readreply(): even if it's not clear if it's possible to receive a response for a query we didn't send, but we better execute auth_dns_callback() immediately instead of waiting until timeout_auth_queries_event() takes action.
- res.c: made timeout_query_list() not require an argument
- Merged Adam's event system rewrite
- Let mp_pool_get() clear memory
- res.c: removed unused definitions
- Renamed MyMalloc() to MyCalloc()
- Create 8.2.x branch
- Moved files: s_user.c -> user.c s_misc.c -> misc.c s_serv.c -> server.c
- Moved irc_res.c to res.c - Moved irc_reslib.c to reslib.c
- irc_res.c: style fixes
- Made irc_nscount an unsigned type
- irc_res.c:make_request(): lowered retries counter to 2
- Style corrections all over the place
- Applied Adam's sendto_one_numeric() changes
- Clean up all files in include/ (fixed indentation, removed whitespaces/tabs) - Fixed copyright years
- Revert to -r1831
- Made all numeric defines use the actual string instead of the numeric value which allows to use gcc's printf format attribute - Remove current message locale implementation
- Replaced all occurrences of ircsprintf with sprintf/snprintf and killed sprintf_irc.(c|h)
- Implemented memory pool allocator which basically is taken from Tor's mempool allocator for Tor cells - Fixed compile warnings in conf_class.c - ./configure --enable-assert works again
- Second time's the charm? Moving svnroot/ircd-hybrid-8 to svnroot/ircd-hybrid/trunk
- irc_reslib.c: add localhost as default nameserver in case we cannot find a suitable entrie in resolv.conf - irc_res.c:res_ourserver(): no need to test against INADDR_ANY. This is a relict of old BIND versions which added INADDR_ANDY as default nameserver if no other servers could be found.
- Get rid of fileio.c. Replace some ircsprintf() with snprintf() while on it
- renaming files: ircd_parser.y -> conf_parser.y ircd_lexer.l -> conf_lexer.l s_conf.c -> conf.c s_conf.h -> conf.h s_log.c -> log.c s_log.h -> log.h
- Rewrite and cleanup half-broken logging subsystem. Logfile rotating is not working yet
- move content of msg.h, ircd_handler.h and handlers.h into parse.h and remove headers accordingly - killed common.h - remove m_killhost.c and m_flags.c from contrib/ - sort out unused header includes here and there
- create ircd-hybrid-8 "branch"
- recreate "trunk"
- replace several ircsprintf with snprintf
- Got rid of irc_addrinfo.c and irc_getnameinfo.c - Fixed broken ipv6 detection due to incorrect use of AC_CHECK_TYPES
- branch off trunk to create 7.3 branch
- move ircd-hybrid-7.2 to trunk
- move list manipulation routines from tools.c to list.c - mem_frob() goes to memory.c - sort out redundant/unneeded header includes
- change resolver/auth to use balloc memory pool
- ircd.c: init_resolver() has to be called *before* read_conf_files(), otherwise
recvfrom() aborts with ENOTSOCK in case there are configured connect{} blocks
with a host that needs to be resolved on the very first configuration read.
- delete_resolver_queries(): remove extranous non-NULL test. ptr->data can never be NULL. let it crash, if it ever happens to be
- remove add_local_domain
- res_readreply(): check for fake replies before doing anything else
- fix possible auth/dns related memleaks
- irc_res.c: fixed wide-spread off-by-one when copying hostnames
- res_readreply, do_query_number: removed ip6.int support. ip6.int delegations were deleted in 2006
- res_readreply(): fix broken logic with ipv6 enabled where we ONLY sent a
second A QUERY in case a -broken- nameserver returns NXDOMAIN
for AAAA when A record exists. fixed and removed workaround
for broken nameservers.
- added a mersenne twister basedn prng
- Cleaned up the resolver some more, added some comments
- Redo NXDOMAIN check so IPV6 still works
- reported by Jon, resolver was not checking for NXDOMAIN as well as SERVFAIL this caused odd DNS reports and took longer to connect. Tested by Jon (ThaPrince). (backported from head)
+ fixed unaligned memory reference in resolver Submitted by: Immo 'FaUl' Wehrenberg
- Removed LocalUser.port - Removed two extra memsets in make_request()
(backported from HEAD) - irc_res.c would send out hundreds? of retry requests on clients that did not resolve, i.e. SERVFAIL. The code was essentailly ignoring SERVFAIL hence generating request after request after request. This appears to fix this major bug. It would be a good idea to add some logging here however.
create 7.2 branch, we can move/rename it as needed.
- svn:keywords
- svn:keywords
- Fix svn:keywords
- imported sources - can be moved later according to the directory/branching scheme, but we need the svn up
This form allows you to request diffs between any two revisions of this file. For each of the two "sides" of the diff, enter a numeric revision.
| svnadmin@ircd-hybrid.org | ViewVC Help |
| Powered by ViewVC 1.1.30 |